Re: [AD] MacOSX build broken?

["Peter Hull" <peterhull90@xxxxxxxxxx>]

I've got some code to fix tls.c. I was meaning to ask Trent about this
(I started, then ran out of time, during November's  Hack Day :P )
Some platforms directly access a struct which has been declared
__thread/__declspec(thread) and some use a function tls_get(), so the
code is duplicated. Could we have everyone using tls_get(), but just
defining it as
static THREAD_LOCAL thread_local_state tls = {
   ...
};

static thread_local_state *tls_get(void)
{
   return &tls;
}
for those platforms that support __thread - this should get optimized
out anyway.

BTW the rest of MacOSX is in a sorry state; I am trying but very slow
progress so far, due to other committments.

Pete

On Dec 13, 2007 6:00 AM, Jon Rafkind <workmin@xxxxxxxxxx> wrote:
>
> Thomas Fjellstrom wrote:
> > On Wed November 28 2007, Thomas Fjellstrom wrote:
> >
> >> On Tue November 27 2007, Jon Rafkind wrote:
> >>
> >>> Thomas Fjellstrom wrote:
> >>>
> >>>> zipup.sh complains pretty loudly once it gets to the part where it
> >>>> generates MacOSX dependencies:
> >>>>
> >>>> Generating MacOS X dependencies...
> >>>> Configuring Allegro for MacOS X ...
> >>>> No text file conversion performed ...
> >>>> Done!
> >>>> gcc -MM -MG -I. -I./include -DSCAN_DEPEND -DALLEGRO_MACOSX src/*.c
> >>>> src/unix/*.c src/misc/*.c src/c/*.c src/compat/*.c demo/*.c >
> >>>> _depend.tmp In file included from
> >>>> ./include/allegro/internal/aintern_events.h:4, from
> >>>> ./include/allegro/internal/aintern_display.h:7, from
> >>>> ./include/allegro/inline/gfx.inl:28,
> >>>>                  from ./include/allegro/gfx.h:453,
> >>>>                  from ./include/allegro.h:45,
> >>>>                  from src/allegro.c:23:
> >>>> /include/allegro/internal/aintern_thread.h:5:10: error: #include
> >>>> expects "FILENAME" or <FILENAME>
> >>>>
> >>>> It seems its been like this since about the 15th of this month, (its
> >>>> broken my auto build script, haven't had a new release since then).
> >>>>
> >>> ill test this out when I get some free time, but which branch were you
> >>> on? looks like 4.9..
> >>>
> >> Yup. 4.2 seems to work ok.
> >>
> >>
> >
> > Any progress on this? My "snapshot" cron job still isn't finishing.
> >
> >
> I fixed things a little bit, but osx is still fairly broken. Things I
> changed thus far
> 1. rename AL_ to ALLEGRO_. I guess this renaming scheme took place in
> other files but osx was missed
> 2. make alosxcfg.h define ALLEGRO_INTERNAL_THREAD_HEADER as "aintuthr.h"
> since osx has pthread support.
> 3. unfortunately I had to define THREAD_LOCAL as nothing for osx because
> osx doesn't respect __thread. I googled around and found a post about
> using pthread_getspecific instead.
> http://lists.apple.com/archives/Xcode-users/2006/Jun/msg00550.html
> 4. commented stuff about 'cursor' in qzmouse.m.
> 5. took out the second argument to _al_event_source_init when
> initializing the keyboard and mouse for osx, which is what the other
> platforms do.
>
> I got a bunch of errors when linking liballeg.dylib with scons and make,
> I didn't have cmake installed though. I'll look into fixing the link
> errors, unless some osx expert can help out.
>
>
> -------------------------------------------------------------------------
> SF.Net email is sponsored by:
> Check out the new SourceForge.net Marketplace.
> It's the best place to buy or sell services
> for just about anything Open Source.
> http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
>
> --
> https://lists.sourceforge.net/lists/listinfo/alleg-developers
>